JPS588024B2 - Detection and cutting device for characters with ruby - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character detection and cutting device, and more particularly to a ruby character detection and cutting device that separates ruby characters from ruby characters.

Traditionally, the Japanese text targeted by Kanji OCR was text without ruby, so it was easy to separate each line or character, but when trying to read general Japanese texts such as novels and reference books. When reading text, ruby is often added to some characters, so in order to read the text at its best, it is necessary to separate the text from the ruby.

However, there are still cases where it is difficult to use conventional detection and extraction methods for Japanese sentences that include ruby characters.

For example, in the ruby ``inaka'' attached to ``inaka'', ``na'' is located between the characters ``田'' and ``sha''.

Therefore, it is not possible to detect the blank space between ``田'' and ``sha'' and separate the two characters.

The present invention has been made in view of the above-mentioned circumstances, and its object is to provide a detection and cutting device for ruby characters that solves the above-mentioned problems.

According to this invention, a photoelectric conversion circuit optically scans a reading form on which Japanese text with ruby characters is printed; and a quantization circuit that quantizes an electrical signal obtained by the photoelectric conversion circuit. , a single-line character buffer circuit that stores a quantized signal corresponding to one line of text in the form, and an end portion of the one-line text stored in this single-line character buffer circuit that is not marked with ruby. a line character end detection circuit that detects the position of the line character end detection circuit; and a character separation position is detected within a predetermined width range from the end detected by the line character end detection circuit of the text in the one line character buffer circuit. The present invention includes a character separation circuit for separating characters, a storage circuit for storing character separation positions detected by the character separation circuit, and a ruby separation circuit for separating characters from ruby according to the character separation positions stored in the storage circuit.

FIG. 1 is a block diagram showing one embodiment of the present invention.

Reference numeral 1 is a document on which Japanese text including characters with ruby is printed, and reference numeral 2 represents an example of one line of characters.

The form 1 is optically scanned by the photoelectric conversion circuit 3, and the character example 2 is converted into an electrical signal.

The electrical signal obtained from the photoelectric conversion circuit 3 is supplied to a binarization circuit 4 and quantized into a binary signal.

This binary signal is stored in a buffer circuit 6 as a binary signal corresponding to one line of character examples via a line separation circuit 5.

That is, the buffer circuit 6 has a capacity to store one line of character strings.

2A to 2D are diagrams showing the input direction of the form 1. FIG.

Figures 2a and 2b show the case where the form 1 is printed in horizontal type, and Figure 2 e+d shows the case where the form 1 is printed in vertical type.

Furthermore, FIGS. 2b and 2d show the case where the form 1 is input in the reverse direction (starting from the end of the sentence).

In either case, the form 1 is shown with the direction of the arrow as above, and the photoelectric conversion circuit 3 is shown with the scanning starting from the upper left end in the right direction.

If the input direction of the form 1 is allowed to be restricted in this way, for example, when a form printed in horizontal writing as shown in FIG. 2b is input in the opposite direction, the buffer circuit 6 will be
This will accommodate reversed character strings, such as .

Converting such a reversed character string into a normal character string as shown in FIG. 3b is called address conversion.

Referring again to FIG. 1, 7 is a format designation circuit which designates whether the input direction of the form 1 is normal or reverse.

This instruction can be given, for example, by the operator looking at the input direction of the form 1 and operating a switch (not shown), or by detecting the orientation of a reference mark provided in advance at an appropriate location on the form 1.

If the format designation circuit 7 indicates the reverse orientation, the contents of the bathopher circuit 6 are converted to the normal orientation via the address conversion circuit 8.

The address conversion circuit 8 has another buffer circuit (not shown) that can accommodate a character string of one line, for example.
This can be realized by sequentially accommodating binary signals taken out sequentially from .

Next, the contents of the buffer circuit 6 or the output of the address conversion circuit 8 are supplied to the bottom of one line detection circuit 9 to detect the bottom of one line of the character string.

Here, the 1-line regular part is the side of the character string where ruby is not added, and represents the position of the last scanning line where the character signal (black) exists when the character string is scanned in the arrangement direction. There is.

As shown in FIG. 4a, the 1-line bottom detection circuit 9 scans the character string 20 sequentially from bottom to top using the scanning lines 21 and determines the number of the scanning line from which the first character signal is obtained. It is determined by detecting the

This information is then supplied to the character separation circuit 10 and used to separate one line of character strings character by character.

As shown in FIG. 4b, the character separation circuit 10 separates character string 2 from
A predetermined scanning range represented by a distance y is sequentially scanned in the vertical direction as shown by a scanning line 22 with the bottom of the first row of 0 as a reference.

The distance y can be set in advance to an appropriate value depending on the size of the characters printed on the form 1.

Projection data of each character is obtained by such vertical scanning, and a position of a predetermined width from the center (character separation position) b
1, b2...bi...bn are determined.

This character separation position data is stored in the memory circuit 11 in FIG.
be accommodated in.

When the character separation position data changes, each character can be easily separated character by character.

For example, if you specify b2 and b3, the binary signal between them will be
It is only a character signal representing .

The ruby separation circuit sequentially takes out two consecutive character separation position data bi and bi+1 from the storage circuit 11 and scans the range defined by them in the row direction.

That is, as shown in FIG. 4C, scanning line 2 starts from the bottom of row 1.
3 to sequentially scan upwards.

By terminating the scanning when no character signal (black) is detected in the scanning line 23 and only a white signal is detected, only the characters excluding the ruby are extracted, and the ruby is separated as shown in Fig. 4d. You can get benefits.

The separated characters obtained by the ruby separation circuit are supplied to the recognition section 14 via the character address conversion circuit 1.3 and are recognized.

Character address conversion circuit 13 is format designation circuit 15
The address of the separator character is converted based on the specification of .

In this address conversion, characters (C or d in FIG. 2) printed in vertical writing as shown in FIG. 5a are rotated by 90 degrees as shown in FIG. 5b.

That is, the character address conversion circuit 13 performs the above-mentioned process when the format designation circuit 7 indicates that the form 1 is in vertical writing mode, and when it is in horizontal writing mode, it outputs separated characters as is.

FIG. 6 is a diagram showing an example of the configuration of the main parts of the embodiment shown in FIG. 1.

When one line of character strings is stored in the buffer circuit 6, binary signals are sequentially taken out in synchronization with the signal CPI output from the scanning circuit 30 and sent to the buffer circuit 33 via the one line black/white determination circuit 31 and the gate 32. supplied to

Below, it is "1" when the binary signal is a character signal (black) and "0" when it is blank (white).

The one-line black-and-white determination circuit 31 outputs an output signal ("1") to terminal A when even one black color is detected in one scanning line indicated by reference numeral 21 in FIG. 4a; An output signal is output to terminal B.

The output signal at terminal A sets flip-flop 35 through gate 34.

Therefore, when scanning of the contents of the buffer circuit 6 is started and black is detected, the gate 32 is opened, and thereafter the contents of the buffer circuit 6 are sequentially stored in the buffer circuit 33.

At this time, the contents of the buffer circuit 33 do not have information below the bottom of the first row.

In FIG. 6, the signal E1 outputted from the scanning circuit 30 is a synchronization signal for each scanning line 23, as shown in FIG. 4a.

On the other hand, the counting circuit 36 counts the number of scanning lines including black,
When the predetermined number of pieces is, for example, 5 or more, an output signal (“1”
) is output.

Assuming that the count value of the counting circuit 36 is 4 and all the next scanning lines are white, an output signal is obtained at the terminal B of the 1-row monochrome determination circuit 31, and the gate 31 is opened.

The output signal of gate 37 is a flip-flop. 35 and the buffer circuit 33 are reset.

This is because if one line of character strings is blank and black is detected due to dirt or the like, the content stored in the buffer circuit 6 is assumed to be blank and the next process cannot proceed.

Now, since the buffer circuit 33 stores data based on the bottom of one line of the character string, the scanning circuit 38 sends a signal CP2 corresponding to the scanning line 22 of width y shown in FIG. , and sequentially sends the data from the buffer circuit 33 to the projection circuit 40. supply

If each scanning line contains black, the projection circuit 40 outputs a signal "
Outputs 1”.

The output signal of the projection circuit 40 is supplied to a character separation position calculation circuit 41 together with a signal E2 that defines the number of scanning lines 22 output from the scanning circuit 38.

The signal is supplied to the character separation position calculation circuit 41.

The character separation position calculation circuit 41 counts the signal E2, and the output signal of the projection circuit 40 changes from "1" to "0" and "0".
The center position A of each character is calculated from the stem values a and b of the signal E2 when the signal E2 changes from "" to "1".

As shown in FIG. 1, a position obtained by adding a predetermined value ΔB determined in advance to this center position A is determined as a character separation position bi.

Note that in FIG. 7, 24 represents the output signal of the projection circuit 4D.

The character separation position data obtained in this way is stored in the storage circuit 42 in sequence.

Next, two consecutive character separation position data b1 and b2 in the storage circuit 42 are set in the first position register 43 and the last position register 44, respectively.

The register 44 is supplied with the synchronizing signal CP3 from the scanning circuit 45, and the number of supplied CP3 is b1.
, b2, an output signal is generated.

The output of the register 43 sets the flip-flop 46, and the output of the register 44 resets the flip-flop 46, so that the output signal of the flip-flop 46 is a mask signal shown at 25 in FIG. 4C.

During the period CP3 defined by this mask signal, the buffer circuit 33 is scanned via the gates 47 and 39, and the white detection circuit 48 and the character buffer circuit 49 are driven.

That is, the data sequentially taken out from the buffer circuit 33 is sequentially stored in the character buffer circuit 49.

On the other hand, the gate 50 is opened by the mask signal, and the signal E3 corresponding to the number of scanning lines 23 from the scanning circuit 45 is supplied to the scanning line number counting circuit 51 to count the number of running lines (2).

The scanning line number counting circuit 51 sets the flip-flop 52 when the count reaches a predetermined value H, and resets the flip-flop 52 when the count reaches H+Δ.

That is, the gate 53 is opened during H-H+Δ.

During this time, if the white detection circuit 48 detects that all of the scanning lines 23 shown in FIG. 4C are white,
The storage of data to the character buffer circuit 49 via gates 53 and 54 is stopped.

Further, when the white detection circuit 48 does not output an output signal during H-H+△, it similarly stops.

Therefore, only the data corresponding to the scanning lines C1 to Cm shown in FIG. 4C are stored in the character buffer circuit 49, and ruby is separated.

As described above, according to the present invention, it is possible to separate the ruby of a character with ruby and the character, and it is possible to greatly expand the characters to be read by a character reading device.

Furthermore, characters with ruby text can be read without changing the configuration of the recognition unit.

Although the above embodiment is configured to extract only the characters excluding the ruby part, it is also possible to detect and cut out the ruby part and supply it to the recognition section.

According to such a configuration, by recognizing ruby characters, for example, the recognition processing of the corresponding characters can be significantly reduced.

It goes without saying that various other modifications can be made.

[Brief explanation of drawings]

Fig. 1 shows an embodiment of the present invention, Fig. 2a,
b, c, d, Figure 3 a, b, Figure 4 a, b,
c, d, FIGS. 5a, b, and 7 are diagrams for explaining the operation of one embodiment of the present invention, and FIG. 6 is a block diagram of the main part of one embodiment of the present invention. 1... Ledger, 3... Photoelectric conversion circuit, 6... Buffer circuit, 9... First line bottom detection circuit, 10... Character separation circuit, 11... Memory circuit, 12... Ruby Separation circuit.

Claims (1)

[Claims] 1. A photoelectric conversion circuit that optically scans a form printed with Japanese text mixed with ruby characters and converts it into an electrical signal, and a quantization circuit that quantizes the electrical signal obtained from this photoelectric conversion circuit. Among the quantized signals obtained from this quantization circuit, a buffer circuit that accommodates a quantized signal corresponding to a character string for one line of the Japanese text, and a position of the bottom of the character string in this buffer circuit are detected. a projection circuit that calculates a projection of a quantized signal within a predetermined width in a direction perpendicular to the arrangement direction of the character strings based on the bottom position detected by this means; a separation position calculation circuit that calculates the separation position of the character string from the projection result; and a ruby circuit that sequentially detects the quantized signal within the range separated by the separation position and separates the ruby and character of the ruby-applied character. 1. A detection and cutting device for characters with ruby, characterized by comprising a separation circuit.